SECURITIES AND EXCHANGE COMMISSION
(Release No. 34-78860; File No. SR-CHX-2016-16)
September 16, 2016
Self-Regulatory Organizations; Chicago Stock Exchange, Inc.; Notice of Filing of Proposed Rule
Change to Adopt the CHX Liquidity Taking Access Delay
Pursuant to Section 19(b)(1) of the Securities Exchange Act of 1934 (“Act”)1, and Rule
19b-42 thereunder, notice is hereby given that on September 6, 2016, the Chicago Stock
Exchange, Inc. (“CHX” or “Exchange”) filed with the Securities and Exchange Commission
(“Commission”) the proposed rule change as described in Items I, II and III below, which Items
have been prepared by the Exchange. The Commission is publishing this notice to solicit
comments on the proposed rule change from interested persons.
I. Self-Regulatory Organization’s Statement of the Terms of Substance of the Proposed
Rule Change
CHX proposes to amend the Rules of the Exchange (“CHX Rules”) to adopt the CHX
Liquidity Taking Access Delay. The text of this proposed rule change is available on the
Exchange’s website at http://www.chx.com/rules/proposed_rules.htm, at the principal office of
the Exchange, and at the Commission’s Public Reference Room.
II. Self-Regulatory Organization’s Statement of the Purpose of, and Statutory
Basis for, the Proposed Rule Change
In its filing with the Commission, the self-regulatory organization included statements
concerning the purpose of, and basis for, the proposed rule change and discussed any comments
it received on the proposed rule change. The text of those statements may be examined at the
places specified in Item IV below. The Exchange has prepared summaries, set forth in sections
A, B and C below, of the most significant parts of such statements.
1 15 U.S.C. 78s(b)(1).
2 17 CFR 240.19b-4.
2
A. Self-Regulatory Organization’s Statement of the Purpose of, and the Statutory
Basis for, the Proposed Rule Change
1. Purpose
Background
The Exchange proposes to adopt the CHX Liquidity Taking Access Delay (“LTAD”).
LTAD is designed to neutralize microsecond speed advantages exploited by low-latency market
participants engaged in latency arbitrage3 strategies that diminish displayed liquidity and impair
price discovery in national market system (“NMS”) securities.4 In sum, LTAD would require all
new incoming orders5 received during the Open Trading State
6 that could immediately execute
3 As used herein, “latency arbitrage” means the practice of exploiting disparities in the
price of a security or related securities that are being traded in different markets by taking
advantage of the time it takes to access and respond to market information. Given its
emphasis on speed, latency arbitrage has resulted in a well-documented and escalating
technology race among certain market participants seeking to obtain ever smaller speed
advantages. See Eric Budish, Peter Cramton and John Shim, “The High-Frequency
Trading Arms Race: Frequent Batch Auctions as a Market Design Response,” Quarterly
Journal of Economics, Vol. 130(4), November 2015 (“Budish Paper”); see also e.g.,
Elaine Wah and Michael Wellman. 2013. “Latency Arbitrage, Market Fragmentation, and
Efficiency: A Two-Market Model.” 14th ACM Conference on Electronic Commerce,
June. In recent years, a significant amount of academic research has been done regarding
the impact of latency arbitrage on the efficiency of securities markets. See id. Many of
these studies have suggested that latency arbitrage exacts a “tax” on liquidity provision
that dissuades liquidity providers from displaying large aggressively priced orders for
fear of their stale orders being taken by latency arbitrageurs before the liquidity providers
have had the chance to adjust such orders when reacting to the same market data. See
Eric Budish, Comment letter regarding “Investors’ Exchange LLC Form 1 Application
(Release No. 34-75925; File No. 10-222)” dated February 5, 2016 (“Budish Letter”).
4 The Exchange notes that while LTAD is designed to neutralize microsecond speed
advantages, liquidity providers would still be required to obtain speed capabilities fast
enough to take advantage of the LTAD.
5 “New incoming orders” are orders received by the Matching System for the first time. As
discussed below, LTAD will not apply to other situations where existing orders or
portions thereof are treated as incoming orders, such as (1) resting orders that are price
slid into a new price point pursuant to the CHX Only Price Sliding or Limit Up-Limit
Down Price Sliding Processes and (2) unexecuted remainders of routed orders released
into the Matching System. See CHX Article 1, Rule 2(b)(1)(C); see also CHX Article 20,
Rule 2A(b); see also CHX Article 20, Rule 8(b)(7). Incidentally, the Exchange is
3
against one or more resting orders on the CHX book, as well as certain related cancel messages,
to be intentionally delayed for 350 microseconds before such delayed messages would be
processed7 by the Matching System.
8 9 All other messages, including liquidity providing orders
(i.e., orders that would not immediately execute against resting orders) and cancel messages for
resting orders, would be immediately processed without delay. LTAD will not delay any
outbound messages or market data.
LTAD is a direct response to recent declines in CHX volume and liquidity in the SPDR
S&P 500 trust exchange-traded fund (“SPY”),10
which the Exchange attributes to latency
proposing to amend CHX Article 20, Rule 8(a)(7), which describes how unexecuted
remainders of routed orders are handled by the Matching System, to delete the word
“new” from the last sentence, so that the rule provides, in pertinent part, that if no balance
exists at the time a part of an unexecuted remainder of a routed order is returned to the
Matching System, it shall be treated an incoming order.
6 See CHX Article 1, Rule 1(qq) defining “Open Trading State.”
7 For ease of reference, “processed” means executing instructions contained in a message,
including, but not limited to, permitting an order to execute within the Matching System
pursuant to the terms of the order or cancelling an existing order, whereas “evaluate”
means the Matching System determining whether a message should be diverted into
LTAD, as described below.
8 The Matching System is an automated order execution system, which is a part of the
Exchange’s “Trading Facilities,” as defined under CHX Article 1, Rule 1(z).
9 As discussed below, the Exchange submits that LTAD is a de minimis intentional access
delay in that it is so short as to not frustrate the purposes of Rule 611 of Regulation NMS
by impairing fair and efficient access to an exchange’s quotations. See Securities
Exchange Act Release No. 78102 (June 17, 2016), 81 FR 40785 (June 23, 2016) (“Final
Interpretation”). Thus, the Exchange’s quotations would continue to be “immediately”
accessible and protected pursuant to Rule 611. See 17 CFR 242.600(b)(3) defining
“automated quotation”; see also 17 CFR 242.600(b)(58) defining “protected quotation”;
see also infra Section 3(b).
10 The Exchange believes that much of the CHX liquidity in SPY and other S&P 500-
correlated securities is provided as part of an arbitrage strategy between CHX and the
futures markets, whereby liquidity providers utilize, among other things, proprietary
algorithms to price and size resting orders on CHX to track index market data from a
derivatives market (e.g., E-Mini S&P traded on the Chicago Mercantile Exchange’s
Globex trading platform). As such, an exchange could not make related adjustments to
4
arbitrage activity in SPY first observed at CHX in January 2016 (“SPY latency arbitrage
activity”).11
Specifically, based on its review of unusual messaging patterns in SPY during the
relevant period, corroborating Participant feedback and analysis of market data,12
the Exchange
believes that SPY latency arbitrage has caused CHX liquidity providers to dramatically reduce
displayed liquidity in SPY (and at times withdraw from the market altogether), which, given
CHX’s significant contribution to overall volume and liquidity in SPY prior to the declines,13
materially decreased liquidity in SPY marketwide, as discussed below.14
The Exchange believes that the best way to minimize the effectiveness of latency
arbitrage strategies on CHX with respect to resting limit orders is to implement an asymmetric
delay, such as LTAD, to deemphasize speed as a key to trading success.15
By delaying liquidity
taking orders, and not delaying liquidity providing orders and related adjustment messages,
these special orders on behalf of liquidity providers pursuant to an order type, such as
pegged orders benchmarked to the NBBO. Compare infra note 16.
11 As discussed in detail under Appendix A below, prior to the beginning of the SPY
latency arbitrage activity in January 2016, CHX volume and liquidity in SPY constituted
a material portion of overall volume and liquidity in SPY marketwide. For example, the
CHX Market Share in SPY as a percentage of Total Volume decreased from 5.73% in
January 2016 to 0.57% in July 2016, while the Control Securities did not experience
similar declines. See infra note 12; see also infra Appendix A; see also infra Appendix B
Calculation Set 1a. Also, the Time-weighted Average CHX Size At The NBBO in SPY
relative to the total NMS Size At The NBBO in SPY decreased from 44.36% in January
2016 to 3.39% of the total NMS Size At The NBBO in SPY in July 2016, while the
Control Securities did not experience similar declines. See infra note 12; see also infra
Appendix A; see also infra Appendix B Calculations Sets 3a and 4a.
12 A detailed analysis (“CHX ETF Analysis”) of the impact of latency arbitrage on
displayed liquidity in SPY at CHX, for the period of August 2015 through July 2016
(“Analysis Period”), may be found under Appendix A. The market data utilized by the
CHX ETF Analysis, as well as defined terms and notes, may be found under Appendix B.
13 See supra note 11.
14 See infra Appendix A.
15 See Mary Jo White, Chair, Securities and Exchange Commission, Speech at Sandler
O’Neil & Partners L.P. Global Exchange and Brokerage Conference (June 5, 2014).
5
LTAD would give liquidity providers a small amount of additional time, the same length as the
Investors Exchange LLC (“IEX”) POP/coil delay (“IEX Delay”) recently approved by the
Commission,16
to cancel or adjust resting orders on the CHX book to comport to the most recent
market data before latency arbitrageurs could take such orders at potentially “stale” prices.17
As
the Commission noted in the IEX Approval Order, a symmetric delay that delays all inbound
messages, such as the IEX Delay, would be ineffective in protecting resting limit orders from
latency arbitrage.18
Thus, the Exchange believes that LTAD will enhance displayed liquidity and
price discovery in NMS securities without adversely affecting the ability of virtually all market
participants, other than latency arbitrageurs, to access liquidity at CHX.19
Additionally, the Exchange notes that adopting a symmetric delay and order types that
would permit the Exchange to reprice resting orders based on undelayed market data (e.g.,
pegged orders), such as the IEX Delay, would not be practical in addressing latency arbitrage
with respect to limit orders because the liquidity provision strategies utilized by CHX liquidity
16 See Securities Exchange Act Release No. 78101 (June 17, 2016), 81 FR 41141 (June 23,
2016) (“IEX Approval Order”). Unlike LTAD, the IEX Delay will delay all inbound
order-related messages from IEX Users, outbound message confirmations to IEX Users,
and outbound market data disseminated through IEX’s proprietary data feed. See IEX
Approval Order at 41154. By not delaying inbound market data, IEX would be able to
reprice its resting pegged orders to track changes to the NBBO before latency
arbitrageurs could execute against such pegged orders at potentially stale prices, which
facilitates the ability of IEX to comply with its rules regarding the repricing of pegged
orders. See IEX Approval Order at 41155.
17 In discussing possible alternatives to a frequent batch auction model for trading
securities, the Budish Paper provides that “the asymmetric delay eliminates sniping and
stops the arms race.” See Budish Paper at 1612.
18 See IEX Approval Order, supra note 16, at 41157.
19 Based on the Exchange’s analysis of cancel activity in SPY at CHX for the period
starting in May 2016 through July 2016, the Exchange believes that if LTAD had been
implemented during that time period, out of a total of 18,316 partially-executed orders in
SPY, 20 liquidity taking orders not attributed to latency arbitrage activity would have not
been executed, a de minimis number in the light of the enhanced liquidity and price
discovery afforded by LTAD. See infra Appendix C.
6
providers in SPY, which provide valuable liquidity to the market overall,20
require cancellations
or adjustments to resting limit orders pursuant to proprietary algorithms held by the CHX
liquidity providers that could not be adequately replicated by CHX.21
In light of the above, the Exchange submits that the proposed rules for LTAD are
designed to operate in a manner that is consistent with the Act in that they are designed to protect
investors and the public interest, are not designed to permit unfair discrimination, and would not
impose any unnecessary or inappropriate burden on competition.22
The Exchange now proposes
the following amendments to the CHX Rules to implement LTAD.
Amended Article 20, Rule 8 (Operation of the CHX Matching System)
Proposed Article 20, Rule 8(h) provides rules that comprehensively describe LTAD.
Specifically, proposed paragraph (h) begins by stating that after initial receipt23
of a new
incoming message, the Matching System will evaluate24
the message to determine whether it is a
“delayable message,” as defined under proposed paragraph (h)(1) below. For the purposes of
such an evaluation only, the Matching System shall not consider Match Trade Prevention
(“MTP”), as described under current Article 1, Rule 2(b)(3)(F).25
If not delayable, the Matching
20 See supra note 12; see also infra Appendices A and B.
21 See supra note 10.
22 See infra Section 3(b).
23 As used herein, “initial receipt” means the time at which the Exchange receives a
message and assigns the message a unique sequence number, which the Exchange utilizes
to determine, among other things, message processing order and ranking on the CHX
book. See CHX Article 20, Rule 8(b).
24 See supra note 7.
25 The purpose of ignoring MTP in LTAD evaluation is to provide a previously delayed
order that would not have triggered MTP an opportunity to execute against the resting
order before the newer incoming order would cancel the resting order after release from
LTAD. The Exchange is proposing unrelated modifications to MTP to contemplate
LTAD, as discussed below.
7
System will immediately process the message without delay. Proposed paragraph (h)(1) defines
“delayable message” and provides that delayable messages shall only include the following:
(A) New incoming orders received during the Open Trading State26
that would take
liquidity from the CHX book.
(B) Cancel and cancel/replace messages for delayed orders that have not yet been
released from LTAD.27
(C) The replace portion of a cancel/replace message where the cancel portion cancels
a resting order and the replace portion would take liquidity from the CHX book.
The Exchange notes that the purpose of delaying the aforementioned cancel and cancel/replace
messages is to minimize gaming opportunities by requiring the delayed order to interact with the
CHX book before it is eligible for cancellation.
Mechanically, upon initial receipt of a new incoming message, the Matching System
would assign the message a unique sequence number, as it does currently, which, in addition to
establishing processing and execution priority, will serve as the starting point for the Fixed
LTAD Period, as described below. The Matching System would then initially evaluate the
message to determine whether it is a delayable message.28
For example, a new incoming limit
26 The Exchange notes that LTAD would not apply during a SNAP Cycle, as described
under CHX Article 18, Rule 1, as orders are not immediately executable at that time.
27 As noted later under proposed paragraph (h), a delayed message may only be delayed
once and, thus, the replace portion of a delayed cancel/replace message shall not be
diverted into LTAD upon release in the event that it would take liquidity from the CHX
book.
28 The Exchange notes that the Matching System processes messages for a given security
serially. Thus, the length of time it takes for a message to be evaluated and/or processed
by the Matching System after initial receipt is herein called “variable message queuing
delay,” as the actual length of the delay depends on the number of precedent messages
that have yet to be evaluated and/or processed by the Matching System and are residing
8
order marked Post Only29
that could not take liquidity from the CHX book would not be a
delayable message because it could not immediately execute against one or more resting orders
on the CHX book. In such a case, the undelayed Post Only order would be immediately
cancelled by the Matching System if it would immediately match with a resting order. Similarly,
a new incoming order marked CHX Only30
that would trade-through a protected quotation of an
external market would not be a delayable message as it would be price slid to a permissible
price.31
However, a new incoming order that could immediately execute against a resting order,
but for the fact that MTP would be triggered and prevent a match, would be considered a
delayable message, as MTP is ignored for the purposes of LTAD evaluation only.32
Proposed paragraph (h) continues by providing that if a message is delayable, the
message will be diverted into the LTAD queue and will remain delayed until it is released for
processing. A delayed message shall become releasable 350 microseconds after initial receipt by
the Exchange (“Fixed LTAD Period”),33
but shall only be processed after the Matching System
has evaluated and processed, if applicable,34
all messages in the security received by the
Exchange during the Fixed LTAD Period for the delayed message. Thus, a message may be
in the “Inbound Queue.” The length of time it takes for a message to be evaluated and/or
processed by the Matching System is herein called “system-processing delay.”
29 See CHX Article 1, Rule 2(b)(1)(D) defining “Post Only.”
30 See CHX Article 1, Rule 2(b)(1)(C) defining “CHX Only.”
31 See CHX Article 20, Rule 5(a)(2).
32 See supra note 25.
33 In the event that then-current messaging volume results in a delayable message being
evaluated after 350 microseconds from initial receipt, the delayable message shall be
diverted into LTAD and be immediately releasable. This will ensure that messages
received during the Fixed LTAD Period for a delayed message are evaluated and
processed, if applicable, before the delayable message is released.
34 For example, an order that could not take liquidity from the CHX book would not be
delayed and would be immediately processed, whereas an order that could take liquidity
from the CHX book would be delayed and would not be immediately processed.
9
delayed for longer than the Fixed LTAD Period depending on the then-current messaging
volume in the security.35
The Matching System will utilize a new market snapshot to process a
released order.36
Also, a delayed message shall retain its original sequence number and may only
be delayed once. In addition, LTAD shall apply to all delayable messages submitted by any
Participant for a security traded on the Exchange that is subject to LTAD. The Exchange may
activate or deactivate LTAD per security with notice to Participants.37
The Exchange also proposes to make corresponding amendments to current Article 20,
Rule 8(d) and (f) to contemplate LTAD. Specifically, the Exchange proposes to add the clause
“subject to paragraph (h) below” at the end of current paragraph (d)(1) so that amended
paragraph (d)(1) provides as follows:
Except for certain orders which shall be executed as described in Rule 8(e), below, an
incoming order shall be matched against one or more resting orders in the Matching
System, in the order in which the resting orders are ranked on the CHX book, pursuant to
Rule 8(b) above, at the Working Price of each resting order, as defined under Article 1,
35 In the event a releasable message is awaiting other messages received during its Fixed
LTAD Period to be evaluated and processed, if applicable, the releasable message would
be subject to an additional unintentional variable delay that is a function of the then-
current messaging volume in the security. See supra note 28; see also supra note 33; see
also infra Examples 1-3.
36 The purpose of a new market snapshot is to ensure that the released order is processed in
a manner consistent with federal securities rules and regulations, such as Regulation
NMS and Regulation SHO.
37 As of the date of this filing, the Exchange anticipates applying LTAD to all securities
traded on CHX. In the event the Exchange decides to activate or deactivate LTAD for
certain securities, the Exchange will communicate the list of securities for which LTAD
will be applied and/or the securities for which LTAD will not be applied, as well as the
effective date(s) of such change(s), through a Customer Service Notification. Any change
to the list of LTAD securities shall not be effective prior to the trading day following the
date of the Customer Service Notification and shall only be effective as of the beginning
of the relevant trading day.
10
Rule 1(pp), for the full amount of shares available at that price, or for the size of the
incoming order, if smaller; subject to paragraph (h) below.
The Exchange also proposes to adopt paragraph (f)(3) to provide that certain cancel messages for
an order in LTAD shall be handled as described under proposed paragraph (h). Incidentally, the
Exchange proposes to replace the semi-colon and the word “and” at the end of current paragraph
(f)(1) with a period.
Moreover, proposed paragraph (h)(2) describes how LTAD would interact with the
Exchange’s current order routing protocol and provides that the portion of a Routable Order38
that is to be routed away, pursuant to current Article 19, Rule 3(a), shall be immediately routed
without delay; provided that the entire unrouted balance of the Routable Order will be diverted
into LTAD upon reaching the price point at which the unrouted balance of the Routable Order
would become a delayable message (i.e., would take liquidity from the CHX book), pursuant to
proposed paragraph (h)(1)(A).
Currently, the Exchange determines where and how to route an order on a price point-by-
price point basis.39
That is, the Exchange does not aggregate all protected quotations and resting
liquidity through multiple price points in making a single order routing decision.40
Thus, to the
extent that an incoming order could take liquidity from the CHX book at a price worse than an
away protected quotation (e.g., incoming sell order at $10.00/share; CHX Best Bid at
$10.00/share and NBB at $10.01/share), the Matching System would not consider the fact that
the incoming order could take liquidity from the CHX book at the time the Matching System is
38 See CHX Article 1, Rule 1(oo).
39 See Exchange Act Release No. 74487 (March 12, 2015), 80 FR 14193 (March 18, 2015)
(SR-CHX-2015-02).
40 See id.
11
evaluating the better priced protected quotation. As such, LTAD may result in a portion of a
Routable Order being immediately routed away and the unrouted remainder being delayed.
Amended Routing Protocol
In light of the possible bifurcation of a Routable Order into an immediately routed
portion and a delayed unrouted portion and the fact that the Exchange does not currently utilize
any Router Feedback41
to augment protected quotations,42
LTAD could result in a single order
being routed twice to satisfy the same protected quotation. In order to eliminate this inefficiency,
the Exchange proposes to amend its current order routing protocol to adopt a single type of
Router Feedback called Immediate Feedback to be applied on an order-by-order basis only.43
Specifically, Immediate Feedback would permit the Exchange’s Routing System to
decrease the number of shares available at an away market by an amount equal to the size of the
immediately routed portion of the Routable Order, on an order-by-order basis, with such
41 “Router Feedback” refers to the use of routed orders (“Feedback Orders”) to augment
protected quotations for the purposes of calculating the NBBO. See Securities Exchange
Act Release No. 74075 (January 15, 2015), 80 FR 3693 (January 23, 2015) (SR-BYX-
2015-03).
42 The consolidated market data disseminated by the securities information processors
(“SIPs”) are the only market data feeds utilized by the Exchange for the handling,
execution and routing of orders, as well as for the regulatory compliance processes
related to those functions. See CHX Article 1, Rule 4. Also, the Exchange does not
currently ignore or modify SIP quote data for away markets under any circumstances
where the SIP data feed shows an uncrossed market. See Exchange Act Release No.
74357 (February 24, 2015), 80 FR 11252 (March 2, 2015) (SR-CHX-2015-01); see also
Securities Exchange Act Release No. 72711 (July 29, 2014), 79 FR 45570 (August 5,
2014) (SR-CHX-2014-10).
43 Bats BYX utilizes three different types of Router Feedback in its calculation of the
NBBO, which includes Immediate Feedback, which is described as follows: “Where
BATS Trading routes an order to a venue with a protected quotation using Smart Order
Routing (a “Feedback Order”), the number of shares available at that the venue is
immediately decreased by the number of shares routed to the venue at the applicable
price level.” See SR-BYX-2015-03, supra note 41, at 3695. Also, all Feedback expires as
soon as: (i) one second passes; (ii) the exchange receives new quote information; or (iii)
the exchange receives updated Feedback information. See id.
12
feedback expiring as soon as: (i) one second passes or (ii) the Exchange receives new quote
information from the away market.44
This would permit the Exchange to utilize Immediate
Feedback to ignore the protected quotation to which the immediately routed portion was routed
when the unrouted delayed portion is released from LTAD, thereby preventing double routing to
satisfy the same protected quotation.45
Examples 1 – 3 illustrate the operation of LTAD. Examples 3 and 4 illustrate the
operation of the proposed amended routing protocol.
Amended Article 1, Rule 2(b)(3)(F) (Match Trade Prevention)
Current Article 1, Rule 2(b)(3)(F) describes the MTP modifier, which prevents matches
between orders that originate from the same MTP Trading Group or MTP sublevel thereunder.46
Also, an order sender must designate one of the following MTP Actions for each order, with the
MTP Action noted on the incoming order controlling the MTP interaction:
MTP Cancel Incoming (“N”): An incoming limit or market order marked “N” will not
execute against opposite side resting interest originating from the same MTP Trading
Group or MTP sublevel, if applicable. Only the incoming order will be cancelled
pursuant to MTP.
MTP Cancel Resting (“O”): An incoming limit or market order marked “O” will not
execute against opposite side resting interest originating from the same MTP Trading
44 Given the length of the Fixed LTAD Period, the Exchange notes that it is unlikely that
Immediate Feedback would expire due to one second passing without new quote
information.
45 Given the length of the Fixed LTAD Period, it is unlikely that the Exchange would
receive a confirmation from the away market prior to the unrouted delayed portion being
released from LTAD.
46 See Securities Exchange Act Release No. 71216 (December 31, 2013), 79 FR 883
(January 7, 2014) (SR-CHX-2013-23); see also Securities Exchange Act Release No.
70948 (November 26, 2013), 78 FR 72731 (December 3, 2013) (SR-CHX-2013-20).
13
Group or MTP sublevel, if applicable. Only the resting order will be cancelled pursuant
to MTP.
MTP Cancel Both (“B”): An incoming limit or market order marked “B” will not execute
against opposite side resting interest originating from the same MTP Trading Group or
MTP sublevel, if applicable. The entire size of both orders will be cancelled pursuant to
MTP.
Given that LTAD may result in newer orders (i.e., orders with lower sequence numbers)
becoming resting orders prior to older orders being released from LTAD,47
the Exchange
proposes to amend current Article 1, Rule 2(b)(3)(F)(iii)(a) and (b), which describe MTP Actions
“N” and “O” respectively, to provide that the newer of the contra-side orders, as opposed to the
incoming order if it is the older order, would be cancelled if the incoming order is marked “N,”
and the older of the contra-side orders, as opposed to the resting order if it is the newer order,
would be cancelled if the incoming order is marked “O.” Moreover, given that a price slid order
that triggers MTP is not always the newer order48
and because the Exchange wishes to maintain
the current handling of MTP when it is triggered by a price slid order, the Exchange proposes to
add clauses to the end of current subparagraphs (a) and (b) that preserve that current handling.
Thus, amended subparagraphs (a) and (b) provide as follows:
(a) MTP Cancel New (“N”): An incoming limit or market order marked “N” will not
execute against opposite side resting interest originating from the same MTP Trading
Group or MTP sublevel, if applicable. Only the newer order will be cancelled pursuant to
47 Currently, a new incoming order that triggers MTP is always newer than the resting
contra-side order. However, LTAD may result in the newer of the contra-side orders
being the resting order and the older order being the incoming order. See infra Example
5.
48 See Example 4 under SR-CHX-2013-20.
14
MTP; provided that the incoming order will be cancelled, even if it is not the newer
order, in the event MTP is triggered by the incoming order being price slid pursuant to
the CHX Only Price Sliding Processes.
(b) MTP Cancel Old (“O”): An incoming limit or market order marked “O” will not
execute against opposite side resting interest originating from the same MTP Trading
Group or MTP sublevel, if applicable. Only the older order will be cancelled pursuant to
MTP; provided that the resting order will be cancelled, even if it is not the older order, in
the event MTP is triggered by the incoming order being price slid pursuant to the CHX
Only Price Sliding Processes.
Example 5 below illustrates how the amended MTP would operate in the context of LTAD.
Examples
The following Examples are illustrative of LTAD and related amendments to existing
functionality, but do not exhaustively depict every possible scenario that may arise under LTAD.
Moreover, the Examples do not necessarily depict the actual technical processes of prioritizing
messages and executing orders.
Example 1: LTAD. Assume that LTAD is operational, all messages are for security XYZ and all
orders are routable. Assume that the system-processing delay49
is 50 microseconds.50
Assume
then at 9:59:59.999999, the NBBO is 10.00 x 10.01, the Inbound Queue and the LTAD queue
are empty and the CHX book is as follows:
49 See supra note 28.
50 The Exchange does not represent that actual system-processing delay is at or near 50
microseconds or that unintentional delays do not exist elsewhere in the Matching System
processes. The figure is being utilized for demonstrative purposes only.
15
Fig 1a: CHX Book
Buy Sell
Empty Order A: 1000 @10.01
Assume then that at 10:00:00.000000, the Exchange receives the following order:
Fig 1b: Inbound Queue
Initial Receipt Message
10:00:00.0000000 Order B: Buy 1000 @ 10.01
Under this Example 1, Order B would be immediately evaluated and diverted into LTAD
because it is a delayable message as it could execute against Order A. Due to the system-
processing delay, Order B would be diverted into LTAD at 10:00:00.000050 and releasable at
10:00:00.000350. The result is that the Inbound Queue would be empty and the LTAD queue
would be as follows:
Fig 1c: LTAD Queue
Releasable Time Message
10:00:00.000350 Order B: Buy 1000 @ 10.01
Example 2: Execution Priority. Assume the same as Example 1 and the NBBO is still 10.00 x
10.01 with CHX being the only market at the NBO. Assume then that the Matching System
receives the following new messages in security XYZ:
16
Fig 2a: Inbound Queue
Initial Receipt Message
10:00:00.000265 Cancel Order A
10:00:00.000305 Order C: Sell 1000 @ 10.02
10:00:00.000310 Order D: Buy 1000 @ 10.01
10:00:00.000325 Cancel Order B
10:00:00.000355 Order E: Sell 1000 @ 10.01
Under this Example 2:
Cancel Order A would be evaluated and processed at 10:00:00.000265 without being
diverted into LTAD as it would cancel a resting order and is not a delayable message.
However, due to the system-processing delay, Order A would actually be cancelled at
10:00:00.000315 and the CHX book would become empty.
Order C would then be evaluated at 10:00:00.000315, due to the variable message queuing
delay,51
and then immediately processed without being diverted into LTAD as it adds
liquidity to the CHX book and it is not a delayable message. However, due to the system-
processing delay, Order C would actually post to the CHX book at 10:00:00.000365 and the
CHX book would be as follows:
Fig 2b: CHX Book
Buy Sell
Empty Order C: 1000 @ 10.02
51 See supra note 28.
17
While Order C was being evaluated and processed by the Matching System, Order B became
releasable from the LTAD queue at 10:00:00.000350. However, given that the Matching
System processes messages serially,52
the Matching System would not consider releasing
Order B until after Order C had been processed at 10:00:00.000365, at which point it would
be handled as follows:
o At 10:00:00.000365, the Matching System would compare the releasable time of
Order B to the initial receipt time of the message at the top of the Inbound Queue:
Order D. Since Order D was received during the Fixed LTAD Period for Order B,
Order D would be evaluated before releasing Order B and immediately processed
without being diverted into LTAD as it adds liquidity to the CHX book and is not a
delayable message. However, due to the system-processing delay, Order D would
actually post to the CHX book at 10:00:00.000415. The result is that the NBBO
would become 10.01 x 10.02 and the CHX book would be as follows:
Fig 2c: CHX Book
Buy Sell
Order D: 1000 @ 10.01 Order C: 1000 @ 10.02
o At 10:00:00.000415, the Matching System would then compare the releasable time of
Order B to the initial receipt time of the next message at the top of the Inbound
Queue: Cancel Order B. Since Cancel Order B was received when Order B was in the
LTAD queue, Cancel Order B would be diverted into LTAD as it is a cancel message
for an order that has yet to be released from LTAD. However, due to the system-
52 See id.
18
processing delay, Cancel Order B would be diverted into LTAD at 10:00:00.000465
and releasable at 10:00:00.000675. The result is that the LTAD queue would be as
follows:
Fig 2d: LTAD Queue
Releasable Time Message
10:00:00.000350 Order B: Buy 1000 @ 10.01
10:00:00.000675 Cancel Order B
o At 10:00:00.000465, the Matching System would then compare the releasable time of
Order B to the initial receipt time of the next message at the top of the Inbound
Queue: Order E. However, given that Order E was received after the Fixed LTAD
Period for Order B had expired, the Matching System would release Order B before
evaluating Order E. Due to the system-processing delay, Order B would actually post
to the CHX book at 10:00:00.000515. Also, given that Order B was initially received
before Order D, Order B would receive execution priority over Order D, pursuant to
Article 20, Rule 8(b)(1). The result is that the CHX book would be as follows:
Fig 2e: CHX Book
Buy Sell
Order B: 1000 @ 10.01 Order C: 1000 @ 10.02
Order D: 1000 @ 10.01
Order E would then be evaluated at 10:00:00.000515, due to the variable message queuing
delay, and since it would execute against Order B, it would be diverted into LTAD at
19
10:00:00.000565, due to the system-processing delay, and releasable at 10:00:00.000705.
The result is that the LTAD queue would be as follows:
Fig 2f: LTAD Queue
Releasable Time Message
10:00:00.000675 Cancel Order B
10:00:00.000705 Order E: Sell 1000 @ 10.01
Cancel Order B would then be released from LTAD at 10:00:00.000675, as there are no
messages received during its Fixed LTAD Period in the Inbound Queue. Thus, Cancel Order
B would be processed and Order B would be cancelled at 10:00:00.000725, due to the
system-processing delay. The result is that the CHX Book and the LTAD queue would be as
follows:
Fig 2g: CHX Book
Buy Sell
Order D: 1000 @ 10.01 Order C: 1000 @ 10.02
Fig 2h: LTAD Queue
Releasable Time Message
10:00:00.000705 Order E: Sell 1000 @ 10.01
Order E would then be released from LTAD at 10:00:00.000725, as the Matching System
was processing Cancel Order B when Order E became releasable at 10:00:00.000705. Order
E would then be processed and fully execute against Order D at $10.01/share at
20
10:00:00.000775, due to the system-processing delay. The result is that the Inbound Queue
and the LTAD queue would be empty and the CHX Book would be as follows:
Fig 2h: CHX Book
Buy Sell
Empty Order C: 1000 @ 10.02
Example 3: Post Only and Routing – Immediate Feedback. Assume the same as Example 2 and
that the NBBO is 10.01 x 10.02 with only one market (“Away Market A1”) displaying 1,000
shares at the NBB (“Protected Bid A1”). Assume also that there are no Protected Bids at $10.00.
Assume then that the Matching System receives the following new messages in security XYZ:
Fig 3a: Inbound Queue
Initial Receipt Message
10:00:00.000800 Cancel Order C
10:00:00.001000 Order F: Buy 1000 @ 10.00
10:00:00.001010 Order G: Sell 2000 @ 9.99
10:00:00.001020 Order H: Sell 2000 @ 9.99
10:00:00.001030 Cancel Order F
10:00:00.001040 Order I: Post Only Buy 1000 @ 10.00
Under this Example 3:
Cancel Order C would be evaluated at 10:00:00.000800 and then immediately processed
without being diverted into LTAD as it would cancel a resting order and is not a
21
delayable message. However, due to the system-processing delay, Order C would
actually be cancelled at 10:00:00.000850 resulting in the CHX Book becoming empty.
Order F would then be evaluated and processed at 10:00:00.001000 without being
diverted into LTAD as it would provide liquidity and is not a delayable message.
However, due to the system-processing delay, Order F would actually post to the CHX
book at 10:00:00.001050. The result is that the CHX Book would be as follows:
Fig 3b: CHX Book
Buy Sell
Order F: 1000 @ 10.00 Empty
Order G would then be evaluated at 10:00:00.001050, due to variable message queuing
delay. Pursuant to the Exchange’s routing protocol, the Exchange would immediately
route 1,000 shares of Order G priced at 10.01/share to satisfy Protected Bid A1.53
Moreover, since the unrouted 1000 shares of Order G could execute against Order F at
10.00, the unrouted 1000 shares of Order G would be diverted into LTAD at
10:00:00.001100, due to system-processing delay, and releasable at 10:00:00.001360.
The result is that the LTAD queue would be as follows:
Fig 3c: LTAD Queue
Releasable Time Message
10:00:00.001360 Order G: Sell 1000 @ 9.99
53 The Exchange notes that the time it takes for the Exchange to receive confirmation from
the away market for a routed order is much longer than the proposed 350 microsecond
LTAD. Thus, it is highly unlikely that the Exchange would receive an execution report
from the away market before a delayed unrouted portion is released from LTAD. See
supra notes 44 and 45.
22
Order H would then be evaluated at 10:00:00.001100, due to variable message queuing
delay. Given that Order H is virtually identical to Order G and that the proposed
Immediate Feedback is only applied on an order-by-order basis, Order H would be
handled exactly as Order G. Specifically, the Exchange would immediately route 1000
shares of Order H priced at 10.01/share to satisfy Protected Bid A1. Moreover, since the
unrouted 1000 shares of Order H could execute against Order F at 10.00, the unrouted
1000 shares of Order H would be diverted into LTAD at 10:00:00.001150, due to system-
processing delay, and releasable at 10:00:00.001370. The result is that the LTAD queue
would be as follows:
Fig 3d: LTAD Queue
Releasable Time Message
10:00:00.001360 Order G: Sell 1000 @ 9.99
10:00:00.001370 Order H: Sell 1000 @ 9.99
Cancel Order F would then be evaluated at 10:00:00.001150, due to variable message
queuing delay, but would be immediately processed without being diverted into LTAD as
it would cancel a resting order and is not a delayable message. However, due to the
system-processing delay, Order F would actually be cancelled at 10:00:00.001200. The
result is that the CHX book would become empty.
Order I would then be evaluated at 10:00:00.001200, due to variable message queuing
delay, but would be immediately processed without being diverted into LTAD as it would
provide liquidity and is not a delayable message. However, due to the system-processing
23
delay, Order I would actually post to the CHX book at 10:00:00.001250. The result is that
the CHX book would be as follows:
Fig 3e: CHX Book
Buy Sell
Order I: Post Only 1000 @ 10.00 Empty
Unrouted remainder of Order G would be released from LTAD at 10:00:00.001360, as all
messages received during the Fixed LTAD Period for Order G have already been
processed.54
Thus, Order G would be processed and given the Immediate Feedback
received from the routed portion of Order G and the fact that the Immediate Feedback
had not expired, the unrouted remainder of Order G would fully execute against Order I
at 10.00/share55
at 10:00:00.001410, due to system-processing delay.56
The result is that
the CHX book would become empty.
Unrouted remainder of Order H would be released from LTAD at 10:00:00.001410 as the
Matching System was processing the unrouted remainder of Order G when the unrouted
remainder of Order H became releasable at 10:00:00.001370. Thus, Order H would be
processed and given the Immediate Feedback received from the routed portion of Order
H and the fact that the Immediate Feedback had not expired, the unrouted remainder of
Order H would post to the CHX book at 10:00:00.001460, due to system-processing
delay. The result is that the CHX book would be as follows:
54 See id.
55 See CHX Article 20, Rule 8(d)(1).
56 The Exchange notes that Order I would receive the liquidity provide credit and Order G
would be charged the liquidity taking fee, pursuant to Section E.1 of the Fee Schedule of
the Exchange, even though Order I was initially received after Order G.
24
Fig 3f: CHX Book
Buy Sell
Empty Order H: 1000 at 9.99
Example 4: Routing – Expired Feedback. Assume the same as Example 3, except that
immediately prior to the unrouted portion of Order G being released, the Exchange received an
updated quote from Away Market A1 displaying 1,000 shares at the $10.01.
Under this Example 4, the Immediate Feedback derived from the immediately routed portion of
Order G would expire and, upon release of the unrouted delayed portion of Order G, the
Matching System would route the entire unrouted portion to satisfy the updated Protected Bid
displayed by Away Market A1.
Similarly, the Immediate Feedback derived from the immediately routed portion of Order H
would also expire and, upon release of the unrouted delayed portion of Order H, the Matching
System would route the entire unrouted portion to satisfy the updated Protected Bid displayed by
Away Market A1.
Example 5: MTP. Assume the same as Example 3, except that Order G and Order I originated
from the same MTP Trading Group and Order G has an MTP Action of “N.”
Under this Example 5, pursuant to the current MTP rules, MTP would be triggered and the
unrouted remainder of Order G would be cancelled, as the current “N” MTP Action requires the
incoming order to be cancelled. However, pursuant to the proposed amended MTP rules, Order I
would be cancelled, as the amended “N” MTP action requires the newer order to be cancelled,
absent a price sliding event.
25
Operative Date
In the event the proposed rule change is approved by the SEC, the proposed rule change
shall be operative pursuant to notice by the Exchange to its Participants. Prior to the operative
date, the Exchange will ensure that policies and procedures are in place to allow Exchange
operations personnel to effectively monitor the operation of LTAD.
Appendix A: CHX ETF Analysis
The purpose of the CHX ETF Analysis is to demonstrate that latency arbitrage activity57
in SPY at CHX (“SPY latency arbitrage activity”) has (1) reduced volume and displayed
liquidity in SPY at CHX and (2) impaired liquidity provision in SPY marketwide. For the
purpose of this CHX ETF Analysis, the following terms shall have the following meanings:58
After Period refers to February 2016 through July 2016.
Analysis Period refers to August 2015 through July 2016.
Before Period refers to August 2015 through December 2015.
Control Average refers to the arithmetic average of a given metric for Control Securities.
Control Securities refers to DIA, IWM, and QQQ.59
Entry Event refers to a trading day in January 2016 on which latency arbitrage activity in
SPY at CHX was first observed.
57 See supra note 3.
58 Other capitalized terms utilized in the CHX ETF Analysis shall have the meanings set
forth under Appendix B.
59 Each of the Control Securities were selected for the following similarities to SPY in that
each is: (1) highly correlated in price movements with a well-known equity market index;
(2) ETFs; (3) traded in CHX's Chicago data center; (4) actively traded in the NMS; and
(5) highly correlated with a futures contract traded electronically on the Globex trading
platform.
26
Entry Month refers to January 2016, the month in which latency arbitrage activity in SPY
at CHX was first observed.
Subject Securities refers to SPY and the Control Securities.
Entry of SPY Latency Arbitrage Activity
During the After Period, the Exchange observed unusual messaging patterns in SPY
whereby executions of large inbound Immediate Or Cancel (“IOC”)60
orders against resting
orders in SPY were frequently followed by the receipt of late cancel messages for the executed
resting orders very soon after the execution. This observation was corroborated by feedback from
liquidity providing Participants that indicated that, unlike prior to the Entry Event, they were no
longer able to reliably cancel or cancel/adjust resting orders on the CHX book in SPY in
response to market changes after the Entry Event. The Exchange believes that each instance of
the unusual messaging pattern is the end result of a race triggered by an away market event (e.g.,
change in market data from a futures market) where the liquidity taker is able to take a resting
order at a stale price before the liquidity provider could adjust the resting order to accurately
reflect the market.61
As such, the SPY latency arbitrage activity has had the following impact on
volume and liquidity in SPY at CHX and away exchanges:
Analysis 1: SPY Latency Arbitrage Activity Reduced CHX Market Share in SPY
Relative to Total Volume in SPY and Disproportionately to Control Securities
As shown under Figure 1, CHX Market Share in SPY as a percentage of Total Volume
dropped by 90.1% from 5.73% in the Entry Month to 0.57% in July 2016, while CHX Market
Share in the Control Average dropped by 45.20% from 5.54% in the Entry Month to 3.03% in
60 See CHX Article 1, Rule 2(d)(4).
61 See supra note 10.
27
July 2016.62
As shown under Figure 2, changes in the average Total Volume during the Analysis
Period for the Subject Securities were highly correlated. Thus, Figure 1 and Figure 2 show that
despite the high correlation between SPY and each of the Control Securities during the Analysis
Period, the CHX Market Share in SPY decreased disproportionately to Total Volume, which the
Exchange submits is attributed to the SPY latency arbitrage activity.
Figure 1. This figure illustrates the decrease in CHX Market Share as a percentage of
Total Volume in the Subject Securities (Index: January 2016=100).63
62 See infra Appendix B Calculation Set 1a.
63 See infra Appendix B Calculation Sets 1a and 1b.
0
20
40
60
80
100
120
Aug2015
Sep2015
Oct2015
Nov2015
Dec2015
Jan2016
Feb2016
Mar2016
Apr2016
May2016
Jun2016
Jul2016
SPY Control Average
28
Figure 2. This figure illustrates the correlation in the Total Volume between SPY and the
Control Average (Index: January 2016 = 100) during the Analysis Period.64
65
Analysis 2: SPY Latency Arbitrage Activity Resulted in Less Aggressively Priced
and Smaller Orders in SPY at CHX
While the Exchange did not observe any discernable change on the NBBO spread in SPY
during the After Period, the Exchange did observe a negative impact on the frequency at which
CHX was at the NBBO in SPY and the frequency at which CHX displayed the largest quote at
the NBBO in SPY during the After Period, while Control Securities experienced either smaller
declines or no declines at all.66
Specifically, the % of Time CHX Was At The NBB decreased from 23.8% in the Entry
Month to 8.2% in July 2016;67
the % of Time CHX Was At The NBO decreased from 23.3% in
64 The correlation coefficients (ρ) over the twelve-month period were: ρ(SPY, DIA) =
0.9118, ρ(SPY, IWM) = 0.8996, ρ(SPY, QQQ) = 0.9392, ρ(SPY, Average) = 0.9493.
65 See infra Appendix B Calculation Sets 2a and 2b.
66 See infra Appendix B Calculation Sets 6 and 7.
67 See infra Appendix B Calculation Set 6a.
0
20
40
60
80
100
120
Aug2015
Sep2015
Oct2015
Nov2015
Dec2015
Jan2016
Feb2016
Mar2016
Apr2016
May2016
Jun2016
Jul2016
SPY Control Average
29
the Entry Month to 5.8% in July 2016;68
and the % of Time CHX Was At The NBB and that
CHX Was At The NBO decreased from 3.3% in the Entry Month to 0% in July 2016.69
Moreover, the % of Time CHX Was At The NBB And Was The Largest Bid At That
Price decreased from 20% in the Entry Month to 2.3% in July 2016;70
the % of Time CHX Was
At The NBO And Was The Largest Offer At That Price decreased from 20.7% in the Entry
Month to 1.1% in July 2016;71
and the % of Time CHX Was At The NBB And Was The Largest
Bid At That Price and that CHX Was At The NBO And Was The Largest Offer At That Price
decreased from 1.9% to 0%.72
These calculation sets clearly show that SPY latency arbitrage activity resulted in less
aggressively priced CHX displayed liquidity in SPY and smaller CHX displayed size at the
NBBO, during the After Period. SPY latency arbitrage also negatively impacted the percentage
of the time that CHX was at the NBBO and the percentage of the time CHX displayed the largest
quote at the NBBO.
Analysis 3: Latency Arbitrage Activity at CHX Reduced CHX Size At The
NBBO in SPY Relative to the Control Securities and NMS Size At The NBBO
As shown under Figure 3, during the Before Period, the Time-weighted Average CHX
Size at The NBBO for SPY tended to follow changes to the Control Average, whereas from the
Entry Month through July 2016, the Time-weighted Average CHX Size At The NBBO for SPY
decreased by 82.16% and the Time-weighted Average CHX Size At The NBBO for the Control
68 See infra Appendix B Calculation Set 6b.
69 See infra Appendix B Calculation Set 6c.
70 See infra Appendix B Calculation Set 7a.
71 See infra Appendix B Calculation Set 7b.
72 See infra Appendix B Calculation Set 7c.
30
Average increased by 64.38%.73
As shown under Figure 4, during the Before Period, the monthly
changes in the Time-weighted Average CHX Size At The NBBO tended to follow similar
changes to the Time-weighted Average NMS Size At The NBBO. However, during the After
Period, the monthly changes in the Time-weighted Average CHX Size At The NBBO in SPY did
not follow changes to the Time-weighted Average NMS Size At The NBBO in SPY. Moreover,
during the After Period, CHX went from having a Two-Sided Market in SPY 100% of regular
trading hours in the Entry Month to 74% of regular trading hours in July 2016.74
Thus, Figure 3 and Figure 4 show that SPY latency arbitrage negatively impacted
liquidity in SPY marketwide. Moreover, the data shows that the change in the risk/reward of
providing liquidity in SPY at CHX which resulted from the introduction of the SPY latency
arbitrage activity resulted in a significant reduction of liquidity in SPY provided by CHX, even
during a period when significant incremental liquidity was being added in the Control Securities.
73 See infra Appendix B Calculation Sets 3a and 3b.
74 See infra Appendix B Calculation Set 5.
0
50
100
150
200
250
Aug2015
Sep2015
Oct2015
Nov2015
Dec2015
Jan2016
Feb2016
Mar2016
Apr2016
May2016
Jun2016
Jul2016
SPY Control Average
31
Figure 3. This figure illustrates the Time-weighted Average CHX Size At The NBBO in
the Subject Securities (Indexed: January 2016 = 100) during the Analysis Period.75
Figure 4. This figure illustrates the Time-weighted Average CHX Size At The NBBO in
SPY versus Time-weighted Average NMS Size At The NBBO in SPY (Indexed: January
2016 = 100) during the Analysis Period.76
Analysis 4: SPY Latency Arbitrage Activity Reduced Displayed Liquidity in SPY
Marketwide
Although the Time-weighted Average NMS Size At The NBBO in SPY increased by
22.83% during the After Period, the increase in SPY did not follow much greater increases in the
75 See infra Appendix B Calculation Sets 3a and 3b.
76 See infra Appendix B Calculation Sets 3b and 4b.
0
50
100
150
200
250
Aug2015
Sep2015
Oct2015
Nov2015
Dec2015
Jan2016
Feb2016
Mar2016
Apr2016
May2016
Jun2016
Jul2016
CHX NMS
32
Time-weighted Average NBBO Size in the Control Group, which increased by 128.82% during
the After Period.77
Moreover, during the After Period, the Time-weighted Average CHX Size At
The NBBO for SPY decreased by 90.61%78
and, as a % of total NMS Size At The NBBO in
SPY, from 44.36% to 3.39%.79
These calculations suggest that the SPY latency arbitrage activity
materially impacted displayed liquidity in SPY marketwide. The dramatic decrease in displayed
liquidity in SPY at CHX during the After Period explains why the increase in Time-weighted
Average NBBO Size in SPY lagged behind the increase in Time-weighted Average NBBO Size
in the Control Securities. Had CHX Size At The NBBO remained at least constant during the
After Period, NBBO Size in SPY would have been at least 32.7% higher in July 2016, as shown
below:80
NMS Size at NBBO Change Attribution
Jan-16 Jul-16 Change CHX Others
SPY 9,513 11,686 2,172 -3,824 5,996
DIA 2,569 4,711 2,142 1,227 915
IWM 5,222 10,026 4,804 536 4,268
QQQ 14,100 35,354 21,253 3,900 17,353
Control Average 7,297 16,697 9,400 1,888 7,512
77 See infra Appendix B Calculation Set 4a.
78 See infra Appendix B Calculation Set 3a.
79 See infra Appendix B Calculations Sets 3a and 4a.
80 See infra Appendix B Calculation Set 4a.
33
Conclusion
Based on its observations of unusual messaging patterns in SPY, feedback from
Participants and the analysis summarized above, the Exchange believes that the unusual
messaging activity in SPY that was first observed in the Entry Month is attributed to SPY latency
arbitrage activity. The market data shows that in response to the SPY latency arbitrage activity,
CHX liquidity providers displayed smaller orders in SPY at less aggressive prices during the
After Period relative to the Before Period and Entry Month. Moreover, in light of CHX’s
significant contribution to overall volume and liquidity in SPY during the Before Period and the
Entry Month, diminished displayed liquidity at CHX has materially impaired displayed liquidity
in SPY marketwide.
Appendix B: Calculation Sets
The calculations sets below were prepared with microsecond-level trade and quote
record. Trade records include the date, microsecond-level timestamp, exchange, security symbol,
price, and quantity of all trades reported to the consolidated tape. Quote records include the date,
microsecond-level timestamp, exchange, security symbol, bid price, bid quantity, ask price, and
ask quantity of all quotes reported to the consolidated tape. Only protected quotations are
reported to the consolidated tape.
The Analysis Period for the calculations begins on August 1, 2015 and ends on July 31,
2016. Symbols SPY and three other Control Securities (i.e., DIA, IWM, and QQQ) were
considered. Only trades and quotes that occurred on the national securities exchanges during the
regular trading hours81
were considered. Certain types of non-standard trades were excluded.82
81 See 17 CFR 242.600(b)(64).
82 Non-standard trades include derivatively priced trades, qualified contingent trades,
opening trades, closing trades, and after hours trades.
34
Quotes with negative prices or quantities were excluded. Unless otherwise indicated, lengths of
time when the market was locked or crossed were not considered.
In the calculations below:
Total Volume refers to the number of shares of the indicated symbol traded on the
national securities exchanges on a given day, excluding certain types of non-standard
trades. CHX Volume refers to the number of shares of the indicated symbol traded on
CHX on a given day, excluding certain types of non-standard trades.
CHX Market Share was calculated as CHX Volume divided by Total Volume on a given
day, CHX Market Share = CHX Volume ÷ Total Volume.
CHX Had A Two-Sided Market refers to an indicator variable defined as true at any
microsecond when there was at least one bid and at least one offer among all outstanding
orders on CHX, and false otherwise. CHX Had A One-Sided Market refers to an
indicator variable defined as true at any microsecond when there was at least one bid but
no offers among all outstanding orders on CHX or when there was at least one offer but
no bids among all outstanding orders on CHX, and false otherwise. CHX Had No Market
refers to an indicator variable defined as true at any microsecond when there were no
outstanding orders on CHX, and false otherwise.
A bid was At The NBB at any microsecond when its price was equal to the National Best
Bid. An offer was At The NBO at any microsecond when its price was equal to the
National Best Offer.
At any microsecond, the NMS Size At The National Best Bid (“NMS Size At The NBB”)
refers to the quantity of shares in prevailing bids on the national securities exchanges
priced at the National Best Bid and the NMS Size At The National Best Offer (“NMS
35
Size At The NBO”) refers to the quantity of shares in prevailing offers on the national
securities exchanges priced at the National Best Offer. NMS Size At The NBBO was
calculated as the average of the National Best Bid Size and the National Best Offer Size
at each microsecond, NMS Size At The NBBO = (NMS Size At The NBB + NMS Size
At The NBO) ÷ 2.
CHX Was At The NBB refers to an indicator variable defined as true at any microsecond
when the CHX Best Bid was at the National Best Bid, and false otherwise. CHX Was At
The NBO refers to an indicator variable defined as true at any microsecond when the
CHX Best Offer was at the National Best Offer, and false otherwise.
At any microsecond, the CHX Size At The NBB (“CHX Size At The NBB”) refers to the
CHX Best Bid Size if CHX was at the NBB and zero if CHX was not at the NBB. At any
microsecond, the CHX Size At The NBO (“CHX Size At The NBO”) refers to the CHX
Best Offer Size if CHX was at the NBO and zero if CHX was not at the NBO. CHX Size
At The NBBO was calculated as the average of the CHX Size At The NBB and CHX
Size At The NBO at each microsecond, CHX Size At The NBBO = (CHX Size At The
NBB + CHX Size At The NBO) ÷ 2.
CHX Was At The NBB And Was The Largest Bid At That Price refers to an indicator
variable defined as true at any microsecond when CHX was at the National Best Bid and
the CHX Best Bid Size was greater than or equal to the largest quantity of shares in
prevailing bids on any one national securities exchange other than CHX, and false
otherwise. CHX Was At The NBO And Was The Largest Offer At That Price refers to an
indicator variable defined as true at any microsecond when CHX was at the National Best
Offer and the CHX Best Offer Size was greater than or equal to the largest quantity of
36
shares in prevailing offers on any one national securities exchange other than CHX, and
false otherwise.
For the calculations in the table below:
• Monthly average values are shown. Monthly average values were calculated as the
average of daily values for each day in a month. Daily values were calculated as time-
weighted averages or as percentages of time in the trading day, as indicated in the table.
Time-weighted average values were calculated as daily average of the specified quantity,
market share, or spread value weighted by time (in microseconds). % of time values were
calculated as the length of time (in microseconds) for which the specified indicator
variable was true divided by the length of time in that trading day, excluding lengths of
time during which the market was locked or crossed or otherwise could not be calculated
(e.g., at the start of the trading day).
37
Symbol
SPY DIA IWM QQQ
Control
Average
No. Calculation Month [1] [2] [3] [4] ([2]:[4])
[1a]
CHX Market
Share (% of
Total
Volume)
Aug 2015 4.32% 3.07% 5.51% 3.40% 3.99%
Sep 2015 6.07% 2.61% 3.82% 3.46% 3.30%
Oct 2015 4.08% 5.95% 2.58% 4.42% 4.32%
Nov 2015 4.49% 8.58% 3.14% 5.13% 5.62%
Dec 2015 4.85% 4.89% 2.53% 4.49% 3.97%
Jan 2016 5.73% 9.13% 3.14% 4.35% 5.54%
Feb 2016 4.78% 9.13% 3.32% 4.41% 5.62%
Mar 2016 2.80% 7.54% 2.38% 3.57% 4.50%
Apr 2016 2.28% 4.41% 2.01% 2.69% 3.04%
May 2016 1.10% 3.53% 2.21% 1.93% 2.55%
Jun 2016 0.90% 5.17% 1.74% 3.00% 3.30%
Jul 2016 0.57% 6.11% 1.22% 1.77% 3.03%
[1b]
CHX Market
Share (% of
Total
Volume)
Index:
January 2016
= 100
Aug 2015 75 34 176 78 72
Sep 2015 106 29 122 80 60
Oct 2015 71 65 82 102 78
Nov 2015 78 94 100 118 101
Dec 2015 85 54 81 103 72
Jan 2016 100 100 100 100 100
Feb 2016 83 100 106 102 102
38
Mar 2016 49 83 76 82 81
Apr 2016 40 48 64 62 55
May 2016 19 39 70 44 46
Jun 2016 16 57 55 69 60
Jul 2016 10 67 39 41 55
Symbol
SPY DIA IWM QQQ
Control
Average
No. Calculation Month [1] [2] [3] [4] ([2]:[4])
[2a]
Average Total
Volume
Aug 2015 130,150,083 6,153,725 26,846,599 33,963,873 23,568,046
Sep 2015 94,627,144 6,552,649 21,381,524 28,452,481 19,947,099
Oct 2015 75,881,581 4,461,519 22,420,310 22,701,556 14,268,977
Nov 2015 63,307,314 3,673,677 16,624,141 17,531,483 10,308,999
Dec 2015 87,011,822 4,969,853 23,287,782 24,474,150 16,211,695
Jan 2016 127,469,871 8,301,912 35,204,822 39,029,308 21,425,674
Feb 2016 97,911,733 6,121,299 27,668,000 35,547,824 18,060,375
Mar 2016 63,333,000 2,521,807 20,709,893 17,600,599 9,724,974
Apr 2016 53,023,531 2,337,084 15,556,074 14,984,599 8,991,216
May 2016 51,578,634 2,016,095 17,899,288 14,856,962 9,822,504
Jun 2016 78,385,026 2,740,421 20,938,721 16,963,513 10,240,678
Jul 2016 49,783,615 2,130,330 14,122,275 11,973,239 5,657,111
[2b] Average Total Aug 2015 102 74 76 87 110
39
Volume
Index: Jan
2016 = 100
Sep 2015 74 79 61 73 93
Oct 2015 60 54 64 58 67
Nov 2015 50 44 47 45 48
Dec 2015 68 60 66 63 76
Jan 2016 100 100 100 100 100
Feb 2016 77 74 79 91 84
Mar 2016 50 30 59 45 45
Apr 2016 42 28 44 38 42
May 2016 40 24 51 38 46
Jun 2016 61 33 59 43 48
Jul 2016 39 26 40 31 26
Symbol
SPY DIA IWM QQQ
Control
Average
No. Calculation Month [1] [2] [3] [4] ([2]:[4])
[3a]
Time-
weighted
Average
CHX Size At
The NBBO
Aug 2015 7,740.13 753.47 2,294.04 3,666.82 2,238.11
Sep 2015 6,217.48 682.18 2,157.29 4,177.88 2,339.12
Oct 2015 7,816.38 1,308.53 2,052.68 6,130.87 3,164.03
Nov 2015 8,983.84 2,439.37 2,158.33 7,182.16 3,926.62
Dec 2015 5,776.73 1,152.21 1,517.59 4,347.08 2,338.96
Jan 2016 4,220.05 1,830.97 1,726.35 4,341.83 2,633.05
Feb 2016 2,642.32 1,829.95 2,004.50 4,523.73 2,786.06
Mar 2016 1,611.90 2,347.82 2,077.08 5,987.78 3,470.89
40
Apr 2016 1,415.95 1,481.35 2,314.10 6,196.84 3,330.76
May 2016 485.23 1,469.69 2,374.66 7,423.33 3,755.89
Jun 2016 565.73 1,772.03 2,188.41 7,994.73 3,985.06
Jul 2016 396.37 3,057.61 2,262.70 8,241.77 4,520.69
[3b]
Time-
weighted
Average
CHX Size At
The NBBO
Index: Jan
2016 = 100
Aug 2015 183 41 133 84 85
Sep 2015 147 37 125 96 89
Oct 2015 185 71 119 141 120
Nov 2015 213 133 125 165 149
Dec 2015 137 63 88 100 89
Jan 2016 100 100 100 100 100
Feb 2016 63 100 116 104 106
Mar 2016 38 128 120 138 132
Apr 2016 34 81 134 143 126
May 2016 11 80 138 171 143
Jun 2016 13 97 127 184 151
Jul 2016 9 167 131 190 172
Symbol
SPY DIA IWM QQQ
Control
Average
No. Calculation Month [1] [2] [3] [4] ([2]:[4])
[4a]
Time-
weighted
Aug 2015 19,257.66 2,609.35 6,511.42 18,471.79 9,197.52
Sep 2015 11,919.38 1,679.93 6,540.46 14,223.92 7,481.44
41
Average
NMS Size At
The NBBO
Oct 2015 18,309.27 2,468.56 6,972.46 19,848.75 9,763.26
Nov 2015 19,257.58 3,930.75 6,963.92 23,442.48 11,445.72
Dec 2015 13,230.66 2,204.20 5,812.28 17,106.74 8,374.40
Jan 2016 9,513.33 2,569.26 5,221.94 14,100.46 7,297.22
Feb 2016 7,417.60 2,489.46 6,340.40 13,869.32 7,566.40
Mar 2016 8,638.39 3,703.26 8,521.28 20,316.43 10,846.99
Apr 2016 9,876.59 3,070.53 9,422.71 23,246.57 11,913.27
May 2016 9,398.26 3,144.93 10,295.88 28,354.88 13,931.90
Jun 2016 9,313.10 3,107.54 9,597.43 28,288.57 13,664.51
Jul 2016 11,685.53 4,711.37 10,026.35 35,353.64 16,697.12
[4b]
Time-
weighted
Average
NMS Size At
The NBBO
Index: Jan
2016 = 100
Aug 2015 202 102 125 131 126
Sep 2015 125 65 125 101 103
Oct 2015 192 96 134 141 134
Nov 2015 202 153 133 166 157
Dec 2015 139 86 111 121 115
Jan 2016 100 100 100 100 100
Feb 2016 78 97 121 98 104
Mar 2016 91 144 163 144 149
Apr 2016 104 120 180 165 163
May 2016 99 122 197 201 191
Jun 2016 98 121 184 201 187
Jul 2016 123 183 192 251 229
42
Symbol
SPY DIA IWM QQQ
Control
Average
No. Calculation Month [1] [2] [3] [4] ([2]:[4])
[5a]
% of Time
CHX Had A
Two-Sided
Market
Aug 2015 99.8% 99.6% 99.7% 99.6% 99.7%
Sep 2015 99.9% 99.9% 99.9% 99.9% 99.9%
Oct 2015 100.0% 99.9% 99.9% 100.0% 99.9%
Nov 2015 99.9% 99.9% 99.5% 99.8% 99.7%
Dec 2015 98.6% 98.3% 98.6% 98.6% 98.5%
Jan 2016 100.0% 99.9% 99.9% 100.0% 99.9%
Feb 2016 99.9% 100.0% 100.0% 100.0% 100.0%
Mar 2016 99.8% 100.0% 100.0% 100.0% 100.0%
Apr 2016 99.3% 99.9% 100.0% 99.8% 99.9%
May 2016 85.2% 99.9% 100.0% 100.0% 100.0%
Jun 2016 73.2% 99.9% 100.0% 100.0% 100.0%
Jul 2016 74.0% 99.9% 100.0% 100.0% 100.0%
[5b]
% of Time
CHX Had A
One-Sided
Market
Aug 2015 0.1% 0.1% 0.0% 0.2% 0.1%
Sep 2015 0.0% 0.0% 0.0% 0.0% 0.0%
Oct 2015 0.0% 0.0% 0.0% 0.0% 0.0%
Nov 2015 0.0% 0.0% 0.0% 0.2% 0.1%
Dec 2015 0.0% 0.3% 0.0% 0.0% 0.1%
Jan 2016 0.0% 0.1% 0.0% 0.0% 0.0%
Feb 2016 0.0% 0.0% 0.0% 0.0% 0.0%
43
Mar 2016 0.2% 0.0% 0.0% 0.0% 0.0%
Apr 2016 0.2% 0.0% 0.0% 0.0% 0.0%
May 2016 3.0% 0.0% 0.0% 0.0% 0.0%
Jun 2016 6.1% 0.0% 0.0% 0.0% 0.0%
Jul 2016 1.8% 0.0% 0.0% 0.0% 0.0%
Symbol
SPY DIA IWM QQQ
Control
Average
No. Calculation Month [1] [2] [3] [4] ([2]:[4])
[5c]
% of Time
CHX Had No
Market
Aug 2015 0.1% 0.3% 0.3% 0.1% 0.2%
Sep 2015 0.0% 0.1% 0.1% 0.0% 0.1%
Oct 2015 0.0% 0.1% 0.1% 0.0% 0.1%
Nov 2015 0.1% 0.1% 0.4% 0.0% 0.2%
Dec 2015 1.4% 1.4% 1.4% 1.4% 1.4%
Jan 2016 0.0% 0.0% 0.0% 0.0% 0.0%
Feb 2016 0.1% 0.0% 0.0% 0.0% 0.0%
Mar 2016 0.0% 0.0% 0.0% 0.0% 0.0%
Apr 2016 0.5% 0.1% 0.0% 0.2% 0.1%
May 2016 11.8% 0.1% 0.0% 0.0% 0.0%
Jun 2016 20.7% 0.1% 0.0% 0.0% 0.0%
Jul 2016 24.2% 0.0% 0.0% 0.0% 0.0%
Symbol
44
SPY DIA IWM QQQ
Control
Average
No. Calculation Month [1] [2] [3] [4] ([2]:[4])
[6a]
% of Time
CHX Was At
The NBB
Aug 2015 16.5% 32.7% 46.9% 58.0% 45.9%
Sep 2015 24.0% 36.4% 44.7% 67.6% 49.6%
Oct 2015 30.8% 45.8% 44.3% 74.9% 55.0%
Nov 2015 24.5% 50.3% 54.0% 79.6% 61.3%
Dec 2015 29.2% 34.1% 38.3% 71.3% 47.9%
Jan 2016 23.8% 46.0% 40.2% 70.4% 52.2%
Feb 2016 15.5% 53.9% 33.7% 65.5% 51.0%
Mar 2016 18.5% 58.4% 35.6% 66.8% 53.6%
Apr 2016 18.7% 46.8% 35.9% 60.5% 47.7%
May 2016 7.0% 44.8% 53.5% 68.5% 55.6%
Jun 2016 5.4% 47.1% 44.2% 72.8% 54.7%
Jul 2016 8.2% 45.9% 40.8% 74.1% 53.6%
[6b]
% of Time
CHX Was
At The
NBO
Aug 2015 27.9% 39.8% 57.0% 65.6% 54.1%
Sep 2015 29.7% 36.0% 41.8% 66.7% 48.2%
Oct 2015 20.9% 41.4% 42.7% 74.0% 52.7%
Nov 2015 28.7% 39.3% 52.9% 78.2% 56.8%
Dec 2015 27.1% 35.5% 42.4% 70.0% 49.3%
Jan 2016 23.3% 52.3% 48.8% 70.4% 57.2%
Feb 2016 23.2% 55.5% 46.3% 69.1% 57.0%
Mar 2016 19.0% 58.5% 44.4% 70.0% 57.7%
45
Apr 2016 14.0% 44.0% 36.4% 65.8% 48.7%
May 2016 12.4% 40.4% 49.3% 64.2% 51.3%
Jun 2016 11.0% 47.3% 48.4% 74.6% 56.8%
Jul 2016 5.8% 46.0% 34.0% 69.4% 49.8%
Symbol
SPY DIA IWM QQQ
Control
Average
No. Calculation Month [1] [2] [3] [4] ([2]:[4])
[6c]
% of Time
CHX Was
At The NBB
and that
CHX Was
At The NBO
Aug 2015 1.0% 8.2% 19.7% 32.5% 20.2%
Sep 2015 2.0% 10.0% 9.2% 37.1% 18.8%
Oct 2015 3.0% 14.4% 10.2% 49.8% 24.8%
Nov 2015 6.0% 14.2% 17.9% 58.1% 30.1%
Dec 2015 4.4% 9.3% 12.5% 44.8% 22.2%
Jan 2016 3.3% 19.2% 7.8% 41.8% 22.9%
Feb 2016 1.0% 24.5% 4.8% 35.4% 21.5%
Mar 2016 0.5% 29.6% 4.6% 38.0% 24.1%
Apr 2016 0.2% 15.7% 2.2% 29.9% 15.9%
May 2016 0.0% 13.5% 17.5% 34.6% 21.9%
Jun 2016 0.0% 17.0% 12.2% 48.5% 25.9%
Jul 2016 0.0% 12.6% 4.0% 44.1% 20.3%
46
Symbol
SPY DIA IWM QQQ
Control
Average
No. Calculation Month [1] [2] [3] [4] ([2]:[4])
[7a]
% of Time
CHX Was At
The NBB
And Was
The Largest
Bid At That
Price
Aug 2015 13.6% 26.2% 37.1% 26.6% 29.9%
Sep 2015 21.5% 34.0% 40.0% 47.6% 40.6%
Oct 2015 24.9% 43.8% 36.2% 57.4% 45.8%
Nov 2015 18.8% 47.9% 39.4% 55.9% 47.7%
Dec 2015 25.1% 31.7% 27.7% 39.1% 32.8%
Jan 2016 20.0% 43.6% 32.0% 48.1% 41.2%
Feb 2016 11.2% 52.7% 28.5% 45.5% 42.2%
Mar 2016 11.9% 55.7% 28.3% 44.8% 42.9%
Apr 2016 13.0% 42.2% 31.6% 43.6% 39.1%
May 2016 1.7% 39.8% 37.9% 50.2% 42.6%
Jun 2016 2.0% 43.7% 32.2% 48.3% 41.4%
Jul 2016 2.3% 43.2% 31.7% 48.0% 41.0%
[7b]
% of Time
CHX Was At
The NBO
And Was
The Largest
Offer At
That Price
Aug 2015 24.3% 34.4% 51.2% 39.8% 41.8%
Sep 2015 27.0% 33.8% 37.8% 46.7% 39.4%
Oct 2015 16.0% 38.1% 31.3% 44.0% 37.8%
Nov 2015 22.6% 36.8% 35.1% 53.4% 41.8%
Dec 2015 23.2% 32.7% 30.6% 36.8% 33.4%
Jan 2016 20.7% 51.1% 41.3% 50.7% 47.7%
Feb 2016 18.5% 54.7% 40.8% 49.4% 48.3%
47
Mar 2016 12.9% 55.2% 35.3% 51.2% 47.2%
Apr 2016 8.1% 38.6% 30.8% 45.9% 38.4%
May 2016 3.8% 36.7% 29.8% 45.2% 37.2%
Jun 2016 4.6% 44.6% 31.4% 51.8% 42.6%
Jul 2016 1.1% 42.5% 27.0% 31.0% 33.5%
Symbol
SPY DIA IWM QQQ
Control
Average
No. Calculation Month [1] [2] [3] [4] ([2]:[4])
[7c]
% of Time
CHX Was At
The NBB
And Was
The Largest
Bid At That
Price and
that CHX
Was At The
NBO And
Was The
Largest Offer
At That Price
Aug 2015 0.2% 5.3% 12.8% 7.1% 8.4%
Sep 2015 1.1% 8.5% 7.3% 16.7% 10.9%
Oct 2015 0.9% 12.3% 5.3% 17.7% 11.8%
Nov 2015 2.3% 12.6% 7.0% 23.0% 14.2%
Dec 2015 2.9% 8.1% 6.4% 13.7% 9.4%
Jan 2016 1.9% 17.3% 4.3% 18.5% 13.4%
Feb 2016 0.3% 23.3% 2.8% 13.9% 13.3%
Mar 2016 0.1% 26.0% 2.6% 14.0% 14.2%
Apr 2016 0.0% 10.9% 1.5% 14.0% 8.8%
May 2016 0.0% 10.4% 8.0% 15.6% 11.3%
Jun 2016 0.0% 14.3% 4.8% 18.6% 12.5%
Jul 2016 0.0% 10.7% 2.8% 10.8% 8.1%
48
Appendix C: Impact of LTAD on Liquidity Takers
The purpose of this analysis is to show that implementation of LTAD would not
materially impact the ability of a random market participant not engaged in a latency arbitrage
strategy to take displayed liquidity at CHX. This analysis assumes that LTAD would not
materially change order sending behavior of Participants.
For the period of May 2016 through July 2016,83
the Exchange observed the following
with regards to SPY:
There were a total of 18,316 orders at least partially executed.
During the same period, the Exchange received 1,278 cancel messages to cancel resting
orders after the resting order had been fully executed (“too-late-to-cancel” or “TLTC”).
Of the 1,278 TLTCs, 412 TLTCs (32.24%) were received sooner than or exactly 350
microseconds after the execution (“TLTC≤ 350”), whereas 866 (67.76%) were received
later than 350 microseconds after the execution (“TLTC> 350”).
Of the 412 TLTC≤ 350, 392 (95.15%) executions were attributed to SPY latency arbitrage
activity while the remaining 20 (4.85%) executions were not.
Of the 866 TLTC> 350, 780 (90.07%) executions were attributed to SPY latency arbitrage
activity while the remaining 86 (9.93%) executions were not.84
Thus, if LTAD had been in effect for the period of May 2016 through July 2016, LTAD
(1) would have prevented up to 412 orders, virtually all of which the Exchange believes were
submitted as part of SPY latency arbitrage activity, from being executed during the 350
83 For the months prior to May 2016 during the Analysis Period, the Exchange did not
maintain TLTC data. A limitation of this data is that CHX Market Share and displayed
liquidity in SPY and, by extension, order sending activity had all diminished considerably
by May 2016. See supra Appendix B Calculation Set 1.
84 See supra note 4.
49
microsecond Fixed LTAD Period and (2) would have had a negative impact on only 20 liquidity
taking orders not attributed to SPY latency arbitrage activity. These 20 orders comprised 0.11%
of the 18,316 orders executed during the period. That is, during the measurement period of 63
trading days, LTAD would have had an adverse effect on approximately one order every three
trading days. Thus, LTAD can make a significant contribution to leveling the playing field
between liquidity providers and latency arbitrageurs with minimal adverse effect on other
liquidity taking orders.
2. Statutory Basis
The Exchange believes that the proposed rule change is consistent with Section 6(b) of
the Act in general,85
and furthers the objectives of Section 6(b)(5) in particular,86
in that it is
designed to promote just and equitable principles of trade, to foster cooperation and coordination
with persons engaged in facilitating transactions in securities, to remove impediments and
perfect the mechanisms of a free and open market, and, in general, to protect investors and the
public interest; and is not designed to permit unfair discrimination between customers, issuers,
brokers, or dealers.
Specifically, the Exchange believes that the proposed rule change would remove
impediments and perfect the mechanisms of a free and open market and, in general, protect
investors and the public interest by enhancing displayed liquidity and price discovery for NMS
securities by minimizing the effectiveness of latency arbitrage strategies that diminish quality
and quantity of liquidity. As shown under the CHX ETF Analysis, latency arbitrage lessens
competition among orders by dissuading liquidity providers from displaying large and
85 15 U.S.C. 78f(b).
86 15 U.S.C. 78f(b)(5).
50
aggressively priced orders, which in turn impairs market efficiency.87
The Commission has
recognized the crucial role that displayed limit orders play in the price discovery process.88
Thus,
the Exchange believes that optimizing liquidity provision on the Exchange will enhance price
discovery for NMS securities and, thereby, enhance market efficiency. To this end, LTAD is
designed to promote displayed liquidity on the Exchange by giving liquidity providers a small
amount of additional time to cancel or adjust orders on the CHX book to comport to the most
recent market data before latency arbitrageurs could take such orders at potentially “stale” prices.
LTAD is designed to achieve these goals without adversely affecting the ability of virtually all
market participants, other than latency arbitrageurs, to access liquidity at CHX.89
Thus, the
Exchange believes that LTAD will encourage liquidity providers to resume posting large
aggressively priced orders on the CHX book, which was their practice prior to the beginning of
the SPY latency arbitrage activity in January 2016, which will enhance liquidity and optimize
price discovery in furtherance of the objectives of Act and in a manner consistent with
Regulation NMS, as described below.
The Exchange also believes that the proposed amendments to the MTP order modifier
would remove impediments and perfect the mechanisms of a free and open market and, in
general, protect investors and the public interest, in that they are designed to avoid certain
87 See Exchange Act Release No. 51808 (June 9, 2005), 70 FR 37496 at 37499 (June 29,
2005) (“Regulation NMS Adopting Release”), which provides, in pertinent part: “To the
extent that competition among orders is lessened, the quality of price discovery for all
sizes of orders can be compromised. Impaired price discovery could cause market prices
to deviate from fundamental values, reduce market depth and liquidity, and create
excessive short-term volatility that is harmful to long-term investors and listed
companies. More broadly, when market prices do not reflect fundamental values,
resources will be misallocated within the economy and economic efficiency – as well as
market efficiency – will be impaired.”
88 See Regulation NMS Adopting Release, id, at 37526.
89 See supra note 19; see also supra Appendix C.
51
unintended consequences of LTAD on the MTP functionality. Specifically, since an order would
be assigned a sequence number prior to being evaluated pursuant to LTAD,90
LTAD may result
in a newer undelayed order being posted to the CHX book before an older delayed order, which
would not otherwise occur today. Under this scenario and assuming that the contra-side orders
trigger MTP and the incoming order is marked “N,” the current MTP rules would require the
incoming older order to be cancelled, whereas the amended MTP handling would require the
resting newer order to be cancelled subject to the exception for CHX Only orders described
under amended Article 1, Rule 2(b)(3)(F)(iii)(a) and (b). Thus, the Exchange believes that the
amended MTP functionality better contemplates LTAD and preserves expected results.
Moreover, the Exchange submits that the proposed rules for LTAD are not designed to
permit unfair discrimination, and would not impose any unnecessary or inappropriate burden on
competition. Rather, by neutralizing speed advantages utilized by latency arbitrageurs, LTAD is
designed to ensure that liquidity providers resume achieving their goals with respect to their
liquidity provision strategies on CHX that, prior to January 2016, resulted in valuable liquidity in
securities such as SPY being provided to the marketplace.91
92
In addition, LTAD would
facilitate the achievement of such goals while having a de minimis impact on random liquidity
takers not engaged in latency arbitrage activities.93
In finding that the rules pertaining to the IEX Delay did not permit unfair discrimination,
and would not impose any unnecessary or inappropriate burden on competition, the Commission
90 See supra note 7.
91 See supra note 11; see also supra Appendix A.
92 Since the Entry Event, the Exchange has observed latency arbitrage activity in other
S&P-correlated securities traded on CHX, which has also negatively impacted displayed
liquidity in those securities.
93 See supra note 19; see also supra Appendix C.
52
recognized that displayed limit orders or non-pegged non-displayed limit orders, the types of
liquidity LTAD is designed to protect, would not benefit from the symmetric IEX Delay94
because the purpose of such limit orders is to post or execute consistent with their fixed limit
price, as opposed to being repriced by an exchange based on changes to the NBBO.95
When also
considering that displayed limit orders and non-pegged non-displayed limit orders -1- are as
vulnerable to latency arbitrage attacks as pegged orders96
and -2- could only be effectively
adjusted by the liquidity provider itself in response to market changes if such orders are provided
as part of a broader liquidity provision strategy that utilizes proprietary algorithms to price and
size such limit orders,97
it logically flows that the best way to protect such valuable displayed
liquidity98
is through an asymmetric delay, such as LTAD, that empowers liquidity providers to
more efficiently execute their liquidity provision strategies that result in valuable displayed
liquidity being provided to the market.99
Thus, given the importance of this displayed liquidity
and the ineffectiveness of symmetric delays in protecting limit orders from latency arbitrage, the
Exchange believes that LTAD is narrowly-tailored to address latency arbitrage as applied to limit
orders and, thus, any discrimination between liquidity providers and liquidity takers is justified
and consistent with the requirements of the Act.100
Further, LTAD will be applied to all
94 See IEX Approval Order, supra note 16, at 41157.
95 See id.
96 See supra note 3.
97 See supra note 10.
98 See supra Appendix A.
99 See supra notes 11 and 12.
100 The Exchange further notes that discrimination between liquidity providers and liquidity
takers, in furtherance of the objectives of the Act, is not without substantial precedence in
the NMS. The Commission has previously approved various initiatives that discriminate
between liquidity providers and liquidity takers. For example, many national securities
53
Participants, thus all Participants that provide liquidity in securities subject to LTAD101
will be
able to benefit from the LTAD.
For similar reasons, the Exchange also believes that the proposed rule change is
consistent with Regulation NMS as LTAD would constitute a de minimis intentional access
delay and is thereby consistent with the requirements of Rule 600(b)(3) of Regulation NMS.102
Moreover, the Exchange further believes that LTAD is consistent with Rule 611103
and Rule
610(d) of Regulation NMS.104
Specifically, the Exchange believes that the proposed rule change is consistent with the
“immedia[cy]” requirement of Rule 600(b)(3) as LTAD is a de minimis intentional access delay
and thereby compatible with the Exchange having an “automated quotation” under Rule
600(b)(3) and thus a “protected quotation” under Rule 611.105
Given that LTAD would enhance
liquidity and optimize price discovery in NMS securities, would apply to all Participants and
would not unfairly discriminate among Participants as it is narrowly tailored to minimize the
effectiveness of latency arbitrage strategies with respect to limit orders, all in furtherance of the
exchanges, including CHX, utilize the “maker/taker” fee model, which discriminates
between liquidity providers and takers for the purpose of incentivizing market
participants to provide liquidity to, and/or take liquidity from, the exchange, depending
on the exchange’s specific implementation. See e.g., Bats BYX Fee Schedule; see also
Section E.1 of the CHX Fee Schedule. Similarly, the CHX offers a Market Data Revenue
Sharing program, whereby only certain liquidity providers could receive a market data
revenue rebate in proportion to the quality of liquidity provided. See Section P.1 of the
CHX Fee Schedule. In fact, the IEX Delay discriminates between liquidity providers with
resting pegged orders and liquidity takers, thereby necessarily discriminating between
liquidity providers that utilize pegged orders and those that do not utilize pegged orders.
101 See supra note 37.
102 See 17 CFR 242.600(b)(3).
103 See 17 CFR 242.611.
104 See 17 CFR 242.610(d).
105 See Final Interpretation, supra note 9, at 40792.
54
objectives of Section 6(b)(5) of the Act, as discussed above, the Exchange believes that LTAD
would not impair fair and efficient access to the Exchange’s protected quotation.106
Moreover, the Exchange believes that LTAD is consistent with the requirements of Rule
611.107
As described above,108
a portion of a Routable Order may be immediately routed away to
execute against away protected quotations, with the unrouted remainder being delayed before
being permitted to execute against an order resting on the CHX book at a price inferior to the
away protected quotations by relying on the proposed Immediate Feedback derived from the
immediate routed portion to ignore the away protected quotation. Given that LTAD is de
minimis in the context of Rule 600(b)(3), it logically flows that LTAD should also be considered
de minimis for the purposes of the “simultaneously routed” Intermarket Sweep Order (“ISO”)
requirement under Rule 611(b)(6). Thus, the Exchange submits that a delay caused by LTAD
between the routing of one or more ISOs to satisfy better priced protected quotation(s) and the
delayed execution of a related order through such protected quotation(s) is consistent with the
requirements of Rule 611(b)(6).
Similarly, a portion of a Routable Order may be immediately routed away to execute
against away protected quotations with the unrouted remainder being delayed before posting to
the CHX book at a price that crosses such away protected quotations. This could result if the
resting order on the CHX book that resulted in the unrouted remainder being delayed was
cancelled before the unrouted remainder were released from LTAD. Under this scenario, given
that LTAD is de minimis in the context of Rule 600(b)(3), it logically flows that the de minimis
delay caused by LTAD between the routing of one or more ISOs to satisfy away protected
106 See id.
107 17 CFR 242.611.
108 See supra Example 3.
55
quotations and the actual display of the related order at a price that crosses such away protected
quotations is permissible and consistent with the requirements of Rule 610(d).109
B. Self-Regulatory Organization’s Statement on Burden on Competition
The Exchange does not believe that the proposed rule change will impose any burden on
competition that is not necessary or appropriate in furtherance of the purposes of the Act. To the
contrary, the Exchange believes that any burden on competition is necessary and appropriate in
furtherance of the purposes of Section 6(b)(5) of the Act because LTAD is functionality that
seeks to enhance liquidity and optimize price discovery by deemphasizing speed as a key to
trading success in order to further serve the interests of investors and thereby removes
impediments and perfects the mechanisms of a free and open market.110
The Exchange further notes that market participants will continue to be able to obtain
CHX book data via the SIPs or through the Exchange’s proprietary book feed, the CHX Book
Feed,111
without delay as the Exchange does not propose to delay any outbound messages or
market data. As such, the Exchange submits that any burden on competition, while necessary and
appropriate in furtherance of the purposes of that Act, has been minimized.
C. Self-Regulatory Organization’s Statement on Comments on the Proposed Rule
Change Received from Members, Participants or Others
No written comments were solicited or received with respect to the proposed rule change.
109 See “Division of Trading and Markets: Responses to Frequency Asked Questions
Concerning Rule 611 and Rule 610 of Regulation NMS.” U.S. Securities and Exchange
Commission, 4 April 2008. Web. 20 June 2016
http://www.sec.gov/divisions/marketreg/nmsfaq610-11.htm (“Question 5.02”); see also
CHX Article 20, Rule 6(c)(3).
110 See supra note 15.
111 See CHX Article 4, Rule 1.
56
III. Date of Effectiveness of the Proposed Rule Change and Timing for Commission Action
Within 45 days of the date of publication of this notice in the Federal Register or within
such longer period (i) as the Commission may designate up to 90 days of such date if it finds
such longer period to be appropriate and publishes its reasons for so finding or (ii) as to which
the self-regulatory organization consents, the Commission will:
A. by order approve or disapprove the proposed rule change, or
B. institute proceedings to determine whether the proposed rule change should
be disapproved.
IV. Solicitation of Comments
Interested persons are invited to submit written data, views, and arguments concerning
the foregoing, including whether the proposed rule change is consistent with the Act. Comments
may be submitted by any of the following methods:
Electronic Comments:
Use the Commission’s Internet comment form (http://www.sec.gov/rules/sro.shtml); or
Send an e-mail to [email protected]. Please include File Number SR-CHX-2016-
16 on the subject line.
Paper Comments:
Send paper comments in triplicate to Secretary, Securities and Exchange Commission,
100 F Street, NE, Washington, DC 20549-1090.
All submissions should refer to File Number SR-CHX-2016-16. This file number should be
included on the subject line if e-mail is used. To help the Commission process and review your
comments more efficiently, please use only one method. The Commission will post all
comments on the Commission’s Internet website (http://www.sec.gov/rules/sro.shtml). Copies
57
of the submission, all subsequent amendments, all written statements with respect to the
proposed rule change that are filed with the Commission, and all written communications
relating to the proposed rule change between the Commission and any person, other than those
that may be withheld from the public in accordance with the provisions of 5 U.S.C. 552, will be
available for website viewing and printing in the Commission’s Public Reference Room, 100 F
Street, NE, Washington, DC 20549, on official business days between the hours of 10:00 a.m.
and 3:00 p.m. Copies of such filing will also be available for inspection and copying at the
principal office of the Exchange. All comments received will be posted without change; the
Commission does not edit personal identifying information from submissions. You should
submit only information that you wish to make available publicly. All submissions should refer
to File Number SR-CHX-2016-16 and should be submitted on or before [insert date 21 days
from publication in the Federal Register].
For the Commission, by the Division of Trading and Markets, pursuant to delegated
authority.112
Robert W. Errett
Deputy Secretary
112
17 CFR 200.30-3(a)(12).
Top Related